Files
derek.holloway 7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
Flatten subprojects into a single base monorepo
2026-09-02 16:34:47 -07:00

164 lines
4.7 KiB
LLVM

; RUN: llc < %s -mtriple=nvptx -mcpu=sm_20 | FileCheck %s --check-prefixes=ALL,G32,LS32
; RUN: llc < %s -mtriple=nvptx64 -mcpu=sm_20 | FileCheck %s --check-prefixes=ALL,G64,LS64
; RUN: llc < %s -mtriple=nvptx64 -mcpu=sm_20 --nvptx-short-ptr | FileCheck %s --check-prefixes=G64,LS32
; RUN: %if ptxas-ptr32 %{ llc < %s -mtriple=nvptx -mcpu=sm_20 | %ptxas-verify %}
; RUN: %if ptxas %{ llc < %s -mtriple=nvptx64 -mcpu=sm_20 | %ptxas-verify %}
; RUN: %if ptxas %{ llc < %s -mtriple=nvptx64 -mcpu=sm_20 --nvptx-short-ptr | %ptxas-verify %}
;; i8
define i8 @ld_global_i8(ptr addrspace(1) %ptr) {
; ALL-LABEL: ld_global_i8
; G32: ld.global.b8 %{{.*}}, [%r{{[0-9]+}}]
; G64: ld.global.b8 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load i8, ptr addrspace(1) %ptr
ret i8 %a
}
define i8 @ld_shared_i8(ptr addrspace(3) %ptr) {
; ALL-LABEL: ld_shared_i8
; LS32: ld.shared.b8 %{{.*}}, [%r{{[0-9]+}}]
; LS64: ld.shared.b8 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load i8, ptr addrspace(3) %ptr
ret i8 %a
}
define i8 @ld_local_i8(ptr addrspace(5) %ptr) {
; ALL-LABEL: ld_local_i8
; LS32: ld.local.b8 %{{.*}}, [%r{{[0-9]+}}]
; LS64: ld.local.b8 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load i8, ptr addrspace(5) %ptr
ret i8 %a
}
;; i16
define i16 @ld_global_i16(ptr addrspace(1) %ptr) {
; ALL-LABEL: ld_global_i16
; G32: ld.global.b16 %{{.*}}, [%r{{[0-9]+}}]
; G64: ld.global.b16 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load i16, ptr addrspace(1) %ptr
ret i16 %a
}
define i16 @ld_shared_i16(ptr addrspace(3) %ptr) {
; ALL-LABEL: ld_shared_i16
; LS32: ld.shared.b16 %{{.*}}, [%r{{[0-9]+}}]
; LS64: ld.shared.b16 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load i16, ptr addrspace(3) %ptr
ret i16 %a
}
define i16 @ld_local_i16(ptr addrspace(5) %ptr) {
; ALL-LABEL: ld_local_i16
; LS32: ld.local.b16 %{{.*}}, [%r{{[0-9]+}}]
; LS64: ld.local.b16 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load i16, ptr addrspace(5) %ptr
ret i16 %a
}
;; i32
define i32 @ld_global_i32(ptr addrspace(1) %ptr) {
; ALL-LABEL: ld_global_i32
; G32: ld.global.b32 %{{.*}}, [%r{{[0-9]+}}]
; G64: ld.global.b32 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load i32, ptr addrspace(1) %ptr
ret i32 %a
}
define i32 @ld_shared_i32(ptr addrspace(3) %ptr) {
; ALL-LABEL: ld_shared_i32
; LS32: ld.shared.b32 %{{.*}}, [%r{{[0-9]+}}]
; LS64: ld.shared.b32 %{{.*}}, [%rd{{[0-9]+}}]
; PTX64: ret
%a = load i32, ptr addrspace(3) %ptr
ret i32 %a
}
define i32 @ld_local_i32(ptr addrspace(5) %ptr) {
; ALL-LABEL: ld_local_i32
; LS32: ld.local.b32 %{{.*}}, [%r{{[0-9]+}}]
; LS64: ld.local.b32 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load i32, ptr addrspace(5) %ptr
ret i32 %a
}
;; i64
define i64 @ld_global_i64(ptr addrspace(1) %ptr) {
; ALL-LABEL: ld_global_i64
; G32: ld.global.b64 %{{.*}}, [%r{{[0-9]+}}]
; G64: ld.global.b64 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load i64, ptr addrspace(1) %ptr
ret i64 %a
}
define i64 @ld_shared_i64(ptr addrspace(3) %ptr) {
; ALL-LABEL: ld_shared_i64
; LS32: ld.shared.b64 %{{.*}}, [%r{{[0-9]+}}]
; LS64: ld.shared.b64 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load i64, ptr addrspace(3) %ptr
ret i64 %a
}
define i64 @ld_local_i64(ptr addrspace(5) %ptr) {
; ALL-LABEL: ld_local_i64
; LS32: ld.local.b64 %{{.*}}, [%r{{[0-9]+}}]
; LS64: ld.local.b64 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load i64, ptr addrspace(5) %ptr
ret i64 %a
}
;; f32
define float @ld_global_f32(ptr addrspace(1) %ptr) {
; ALL-LABEL: ld_global_f32
; G32: ld.global.b32 %{{.*}}, [%r{{[0-9]+}}]
; G64: ld.global.b32 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load float, ptr addrspace(1) %ptr
ret float %a
}
define float @ld_shared_f32(ptr addrspace(3) %ptr) {
; ALL-LABEL: ld_shared_f32
; LS32: ld.shared.b32 %{{.*}}, [%r{{[0-9]+}}]
; LS64: ld.shared.b32 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load float, ptr addrspace(3) %ptr
ret float %a
}
define float @ld_local_f32(ptr addrspace(5) %ptr) {
; ALL-LABEL: ld_local_f32
; LS32: ld.local.b32 %{{.*}}, [%r{{[0-9]+}}]
; LS64: ld.local.b32 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load float, ptr addrspace(5) %ptr
ret float %a
}
;; f64
define double @ld_global_f64(ptr addrspace(1) %ptr) {
; ALL-LABEL: ld_global_f64
; G32: ld.global.b64 %{{.*}}, [%r{{[0-9]+}}]
; G64: ld.global.b64 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load double, ptr addrspace(1) %ptr
ret double %a
}
define double @ld_shared_f64(ptr addrspace(3) %ptr) {
; ALL-LABEL: ld_shared_f64
; LS32: ld.shared.b64 %{{.*}}, [%r{{[0-9]+}}]
; LS64: ld.shared.b64 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load double, ptr addrspace(3) %ptr
ret double %a
}
define double @ld_local_f64(ptr addrspace(5) %ptr) {
; ALL-LABEL: ld_local_f64
; LS32: ld.local.b64 %{{.*}}, [%r{{[0-9]+}}]
; LS64: ld.local.b64 %{{.*}}, [%rd{{[0-9]+}}]
; ALL: ret
%a = load double, ptr addrspace(5) %ptr
ret double %a
}